基于动态规划的简单语义单元词义消歧

Word sense disambiguation for simple semantic units based on dynamic programming

为了更为高效、准确的使用语义信息来进行自然语言处理,提出了一种基于动态规划的简单语义单元词义消歧方法。阐述了语义相关度计算模型,提出简单语义单元的概念,分析了简单语义单元的特点;使用动态规划的方法,依次求出每个多义词所有义项的最大生成树变量;求出语义修饰关系完全图的最大生成树,可将最大生成树转化为最佳语法分析方案,实现词义消歧。实验结果表明,该方法具有一定的可行性。

Abstract： To use semantics more effectively and accurately in natural language processing, a word sense disambiguation algo- rithm for the simple semantic units based on dynamic programming is proposed. Firstly the semantics computing model base on semantic relevancy is described, and the definition of the simple semantic units is given, and the characteristics of the simple se- mantic units are analyzed. Using dynamic programming method, the maximum spanning tree variables for each sense of every polysemy are calculated successively, and the maximum spanning tree of the complete semantic modification graph is obtained. The best parsing method is easily transformed from the maximum spanning tree and the word sense of the polysemy is disambiguated with the maximum spanning tree. Finally some experiments are finished to verify the effeteness of the algorithm.